Part 6/11:
In terms of carbon emissions, a 2019 study estimates that training a single large-scale AI model can produce over 626,000 pounds of CO2—a similar footprint to that of an average passenger car over five lifetimes. However, day-to-day usage emissions eclipse training emissions, indicating a systemic energy consumption issue; approximately 80% of total AI-related emissions come from usage, with just 20% from training the models.
This is crucial since diverse AI applications yield different energy costs; for instance, generating images with AI demands significantly more energy than producing text. The astronomical number of images processed daily, coupled with AI's ongoing integration into social media, escalates the total impact substantially.